Talk:Drum memory

Drum memory ‎ (need citation for invention)

This links to US patent 1880523 http://history-computer.com/ModernComputer/Basis/hard_disk.html

Viralmeme (talk) 12:14, 12 August 2010 (UTC)

Dispute universality of head-per-track

While the overwhelming majority of magnetic drums had a head per track, the Sperry Rand UNIVAC FASTRAND had moving heads. Shmuel (Seymour J.) Metz Username:Chatul (talk) 10:43, 18 June 2010 (UTC)

perhaps you would like to edit the article rather than just tag it and make someone else do the work ?? It would have taken less time and effort to just insert the information than it would to write all this and the tag

Chaosdruid (talk) 10:56, 18 June 2010 (UTC)

I already have it on my todo list to update it if nobody beats me to it. It will probably take considerably more time than the brief note on the talk page.

FWIW, I come from a systems programming background where things had to be checked and double checked; the be bold dictum does not come naturally to me in this context. :-( Shmuel (Seymour J.) Metz Username:Chatul (talk) 11:37, 18 June 2010 (UTC)

Why remove reference to fixed-head disks as alternatives to drums?

The 16:54, 12 August 2010 update by Tom94022 removed the text , and fixed-head disks with one head for each track were commonplace in the 1960s, . Why was this done and why was it not reflected in the change description (As noted in the intro, the principal replacement technology for drums was magnetic core.)? Shmuel (Seymour J.) Metz Username:Chatul (talk) 22:33, 12 August 2010 (UTC)

The sentence as constructed was a bit of a non-sequitor so I fixed it by removing the drums vs fixed head drives and focusing on the moving head drums. On the whole I don't think FH HDDs were particularly important but I have no objection to adding a separate paragraph. It should note that both drums and FH HDDs as paging devices were ultimately displaced by improvements in HDD performance and caching at the controller. The problem is this takes a bit of work to find a reliable source for such comments. Tom94022 (talk) 00:23, 13 August 2010 (UTC)

I'd say that the drop in price for main memory was the main factor in making fixed-head rotating DASD unattractive. I've added a new paragraph, which in keeping with the original text does not have references to manuals. I could add references if you consider it appropriate, but thought that it might be TMI. Shmuel (Seymour J.) Metz Username:Chatul (talk) 13:22, 15 August 2010 (UTC)

The 2305 was never enhanced but it wasn't withdrawn until 1980 and my friends and STK tell me their 4305 Solid State Cache had a good run from 1978 until the middle 80s at which time large addressable low cost memories and corresponding OS software changes eliminated the demand for paging files, with the software change being far more important. About the same time caching storage control units became common. Since neither the 2305 nor the 4305 were high volume units, it seems to me that most paging was on high performance HDDs and caching subsystems and not FH HDDs.

I think "commonplace" overstates the significance and changed the term. Tom94022 (talk) 16:47, 15 August 2010 (UTC)

Storage Capacity

How many bits did the first units store? What was the maximum capacity at the time of obsolescence ? What was the transfer rate and rotation speed? --70.167.58.6 (talk) 23:03, 24 January 2011 (UTC)

There were a lot of vendors using drum memories in the 1950's, and the best that you're likely to see is cpacaity and performance information on a sample of those. The IBM 650 that I worked on had a 2,000 word (10 digits plus sign) drum, but I don't know how it compared to earlier drum units.

Similarly, I would expect that you won't get a definitive answer on the most recent capacities and speeds. The most recent drums that I am aware of are the IBM 2301, the IBM 2303, the UNIVAC FASTRAND and the UNIVAC Flying Head Drum; you should be able to find data on those in bitsavers Shmuel (Seymour J.) Metz Username:Chatul (talk) 20:32, 7 February 2011 (UTC)

I don't know if there are bigger ones, but one substantial size drum on a system I used is the one on the Electrologica X8 system at TU Eindhoven, around 1970. Its capacity was 512k 27-bit words (word addressable even though it was secondary storage, so the claim that secondary storage drums are block-addressed is not uniformly true). Paul Koning (talk) 13:30, 15 April 2024 (UTC)

I just checked the UNIVAC 1108 (36-bit word) documentation at bitsavers; the Flying Head drums went up to 2M words and the FastRand drums went up to 33M words per unit. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 15:32, 15 April 2024 (UTC)

Development History

Can anyone clarify the role of Andrew Donald Booth in the development of drum memory? On his own Wikipedia page, he's described, in part, as "a British electrical engineer, physicist and computer scientist who led the invention of the magnetic drum memory for computers." Andrew_Donald_Booth This appears to be at odds with the statement in this article that "Drum memory was a magnetic data storage device invented by Gustav Tauschek in 1932 in Austria." Userboy87 (talk) 16:37, 27 May 2014 (UTC)

Given that Booth was in college in 1937 it is pretty clear that Tauschek's 1932 invention has precedence. According to one Booth bio, his work on a drum began after his US visit which ended Sept 1947. Magnetic Recording - the First 100 years by Daniel et al has drums being developed in the US during WW II and the work continuing at ERA into 1946 and 1947. Thus it seems clear that Booth did not invent the drum although it is likely he had no knowledge of either Tauschek or ERA. I'm going to change the Booth citation. Tom94022 (talk) 06:02, 28 May 2014 (UTC)

See also Tauschek's patent Tom94022 (talk) 06:37, 28 May 2014 (UTC)

Other than it not being IBM, why remove the 1948 Birkbeck work? Certainly "Not notable" is no reason to remove a section from an article - please see WP:N. Andy Dingley (talk) 16:27, 24 July 2017 (UTC)

Aren't patents too primary a source for us? (You can patent anything, even if you never build any.) Does anyone else discuss Tauschek's invention, did he sell any drum memories? --Wtshymanski (talk) 16:49, 24 July 2017 (UTC)

This old issue was recently resurrected in the article. I'm not sure prototype history is notable but if so the then the ERA is the appropriate edit. To answer @Andy_Dingley, Booth's prototype was one of many which did not lead to a product and is simply not notable. Tom94022 (talk) 16:47, 24 July 2017 (UTC)

Welcome to Wikipedia, our standards for notability are admittedly confusing, but you ought to be looking at WP:UNDUE for this issue, not WP:NOTABLE. Andy Dingley (talk) 16:52, 24 July 2017 (UTC)

Thanks, I should have said WP:UNDUE. Can I take it that u agree that Birkbeck is undue? Tom94022 (talk) 21:02, 24 July 2017 (UTC)

No, of course not! It's a 1940s computer. Almost anything that qualifies as a "1940s computer" is rare enough to be notable, let alone passing UNDUE. In particular, this was the work of Andrew Booth, who is credited as the developer of the magnetic drum memory in the UK, and frequently claimed (incorrectly, as it ignores the pre-war work outside the UK) to be the first developer of drum memory at all. In particular, I'd like to see this expanded to cover Booth's influence on the GPO and the first Register-Translator sets (RT4 in 1958) to use magnetic drum memory.

Why would you not include Booth here? Andy Dingley (talk) 21:56, 24 July 2017 (UTC)

Perhaps the question is how many prototypes (failed and otherwise) would you include in this article beyond the invention and the first production unit? US Navy/ERA appears to be the first prototype and it led to a production unit while the IBM 650 appears to be the first production unit. IMO anything else is cummulative, undue and/or not encyclopedic. I don't disagree that Booth's work is interesting and could be expanded, perhaps in his article, but not in this article. Tom94022 (talk) 23:26, 24 July 2017 (UTC)

We've got full articles on the Zuse machines, including the wacky Z1 (V1?). We've got "articles" (so-called) on individual people who's only claim to notability is playing in a basketball tournament. We're hardly likely to have this article overswamped by 1940s-1960s computing machinery, there weren't so many using drums back then that the article will become unmanageable. After all this was not the era of "different laptops every 18 months at Staples", any one of these pre-microprocessor computers is going to have a notability unlike any number of smartphones or tablets.--Wtshymanski (talk) 17:30, 25 July 2017 (UTC)

To quote WP:PROPORTION "An article should not give undue weight to minor aspects of its subject, but should strive to treat each aspect with a weight proportional to its treatment in the body of reliable, published material on the subject." Just because other articles don't comport is no justification for this article giving undue recognition of the Birkbeck drum, certainly a minor aspect of the history of magnetic drums. Tom94022 (talk) 00:48, 26 July 2017 (UTC)

BTW Mee an expert in the history of drums (actually all magnetic recording) does not mention the Birkbeck drum so perhaps we should consider citing Birkbeck as undue in that context. For example Mee cites the Harvard Mark III finished in September 1949 and delivered to the U.S. Navy in March 1950 used nine drums but is not mentioned herein. It is contemperaneous with Birkbeck and unlike Birkbeck it was delivered and used. We could swamp this article with the list of every drum used in a computer from 1940 to 1960 but that would IMO be not very encyclopedic. There has to be a line and if we can't come to a concensus then I guess I will have to add more starting with the Harvard Mark III. Tom94022 (talk) 02:56, 26 July 2017 (UTC)

Go for it! I, for one, would consider that list of much more encyclopediac value than, say, our list of non-existent TTL chips. --Wtshymanski (talk) 03:53, 26 July 2017 (UTC)

Units for stating capacity

While modern readers are familiar with capacities stated in 8-bit bytes, giving capacity in bytes can be misleading for older devices. These were normally organized either by 6-bit characters or by words. Further, some decimal machines used two decimal digits to represent a character. As an example, the IBM 650 could store 2,000 words of 10 digits, and used 2 digits for each word, giving 10,000 characters, not the 8.5K naïvely suggested by the capacity in bytes.

I suggest giving capacities in the native units and approximately equivalent bytes, with a one-time note not to read too much into the latter. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 14:00, 28 February 2022 (UTC)

Agree and changed using straight mathematical conversion or words to bits and then dividing by 8,000 to get kilobytes and ignoring information content. However, I am not sure which set of units should be displayed. I edited in the form of native followed by current equivalent displayed. For the modern readers of this article perhaps it would read better if the only units shown would be current equivalent bits, bytes or kilobytes with the native units in a note or footnote. Tom94022 (talk) 19:37, 7 March 2022 (UTC)

For text, what matters is the number of characters in the native format, the number of six-bit characters is greater than the normalized size in 8-bit bytes. For numeric data it's a bit dicey, but to a first approximation what counts is the number of words in the native format. I'd recommend giving the capacity in native units with a gloss giving an equivalent in 8-bit bytes. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 19:58, 7 March 2022 (UTC)

Notes

650 memory size

@Tom94022: The drum on the 650 had a capacity of 1,000, 2,000 or 4,000 10-digit words plus unit-record buffers. The required control unit for tape and disk had an additional 60 words, but that was core, not drum. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 21:48, 7 March 2022 (UTC)

True, but the original was either 1k or 2k so up to 2k is correct and that it was doubled to 4k is also correct - I suggest more detail is inappropriate for this article and belongs in the IBM 650 article Tom94022 (talk) 01:53, 8 March 2022 (UTC)

Sorry, I misinterpreted your reference to 2 sizes as referring to the product life rather than to the initial introduction, which is why I listed all three. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 11:18, 8 March 2022 (UTC)

Cylinder as logical unit--terminology?

It's mentioned that sometimes drum unit address space was divided into logical cylinders. I am curious if this is accurate contemporary terminology or linguistic "back-porting" from hard disk terminology? In an HDD, a logical cylinder is a stack of tracks through every platter in the drive, forming a theoretical hollow cylinder or tube. If this paradigm were applied to a drum, the result would be a one-item array of a single cylinder comprising every track on the disk---seemingly not very useful as a logical unit. Okto8 (talk) 15:07, 27 October 2022 (UTC)

In IBM S/360 drums (e.g. IBM 2301 are treated as DASD and addressed logical cylinder head record, see Count key data. There were moving head drums, e.g. Fastrand that arguably would support a cylinder concept but AFAICT the term was not used that way by Univac. So it is accurate accurate contemporary terminology. Tom94022 (talk) 17:49, 27 October 2022 (UTC)

Had the UNIVAC division of Remington Rand manufactured a moving-head drum with multiple R/W heads on the access mechanism, logical cylinders might have made sense. I'm not aware of such a device. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 05:50, 28 October 2022 (UTC)

What you've described is a physical cylinder; a logical cylinder is a group of tracks not defined by geometry. The IBM 2301 and 2303 had logical cylinders. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 05:50, 28 October 2022 (UTC)

The size?

Fascinating piece of history, now largely forgotten. But what are the dimensions of these devices? 217.27.182.142 (talk) 17:43, 1 January 2023 (UTC)

Is use of the term drum machine really obvious?

@EEng: A recent edit by ‎EEng removed the sentence It was so common that these computers were often referred to as drum machines. with the comment trivial and obvious. I don't believe that it is at all obvious that these machines were referred to by their memory technology. Shmuel (Seymour J.) Metz Username:Chatul (talk) 01:34, 19 February 2023 (UTC)

Of course it's obvious: adjective noun. Our article on gasoline engines says Applications of petrol engines include automobiles, motorcycles, aircraft, motorboats and ... lawn mowers, chainsaws and portable generators. It does not go on to say, "Such products were so common that they were often referred to as gasoline-engine automobiles, gasoline-engine motorcycles, gasoline-engine aircraft, gasoline-engine motorboats, gasoline-engine lawn mowers, gasoline-engine chainsaws and gasoline-engine portable generators," because it would be stupid to insult the reader's intelligence that way.

But let's say you don't grasp that. How about this: stuff articles say needs to be cited to a source giving that information. But the assertion you quote above is cited to a source that says, simply, For Bendix and Ramo-Wooldridge, the G-20 and RW-400 were parallel core machines rather than serial drum machines of the type already in their product lines. There's nothing in there about anything being "common" or anything being "often referred to" in any particular way. It's just someone using the term drum machine once. You need a source to back up those statements -- but I seriously doubt you'll find one because (to give another example) it's just too stupid to rub the reader's face into the idea that a locomotive might be steam-powered by telling him that "locomotives run by steam were so common that they were often referred to as steam locomotives".

Actually the examples EEng cites suggest that drum machine is an appropriate term. The terms "gasoline-engine automobiles, etc" are rarely if ever used while we do have "diesel locomotives" and "steam locomotives." "Drum machine" was used and is a term of the art as shown by EEng example so it is appropriate for this article, without necessarily saying the term is obvious. So I am going to put "Drum machine" back in albeit not pedantically. Tom94022 (talk) 07:19, 19 February 2023 (UTC)

Nobody ever suggested that it's not an appropriate term. My cavil was with the strained explanation that there was a name for them because a name was needed for them. So I'm glad that, at least, we're not telling the reader that. Unfortunately, you're still using a source giving two examples to support the statement that Many early computers ... used drum memory as the main working memory. What you need is a source saying what the article says. EEng 01:08, 21 February 2023 (UTC)

Claiming that it is obvious does not make it so, and neither do bogus analogies and ad hominem arguments. Stick to the facts and don't make foolish assumptions about who does not grasp what. Please read WP:AGF, WP:CIVIL and WP:FETCH. It is by no means obvious whether a given term is, was or will be used int the literature, and mentioning actual practice does not in any way rub the reader's face into anything; the claim is not that the term might be used, but rather that it actually was. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 13:19, 19 February 2023 (UTC)

You used a single example of actual practice as a source for the idea that they were "so common". Sorry, 1 example <> common. EEng 01:08, 21 February 2023 (UTC)

For large values of one. However, feel free to reword the text to indicate that the term was used without saying how prevalent it was. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 16:26, 21 February 2023 (UTC)

Can we stop this dialog - see ngram on drum machine, drum computerTom94022 (talk) 17:37, 21 February 2023 (UTC)

References